摘要
为实现对微米长纤维模压汽车构件的产品性能进行预测,建立了模压产品性能预测神经网络模型。根据对工艺和实验数据的分析及预测目标等条件,确定了模型的结构。通过对比不同学习算法的性能,选择LM算法为模型的学习算法。以实验数据为样本训练并测试所建模型。试验结果表明:所建预测模型具有较高的预测精度,基本满足产品性能预测的要求。
In order to realize the performance forecast of micron length wood fiber mould pressing automotive component, a neural network model of mould pressing product was established to forecast the performance of automotive component. The model structure was designed based on the analysis of technology and experiment data as well as forecast target. Levenberg- Marquardt(LM) algorithm was selected as the neural network model through comparing the performances of different study algorithms. Then the established model was tested with experimental data. Result indicates that the forecast model has high forecasting accuracy, and it can satisfy the requirements of performance forecast.
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2008年第10期86-87,共2页
Journal of Northeast Forestry University
基金
国家"948"项目(2005-4-62)
关键词
BP网络
微米长纤维
模压制品
性能预测
LM算法
BP network
Micron length wood fiber
Mould pressing product
Performance forecast
Levenberg-Mar-quardt (LM) algorithm